[hinkley]

We talk about this at least once a week here, usually in the scope of software, but it’s a problem that exists in other domains (though VCs and us are constantly pushing for more businesses to run like software, which is nuts considering how not in order we know our own house to be).

Ageism and cultishness are how you end up with simple oversights that anyone who has been around the block a few times would either prevent, or at least have enough self consciousness not to breathlessly talk about it in front of strangers.

I think what we miss consistently about the VC model is that we make the 10:1 ratio about chance and betting big, but I suspect the dominant force there is that 4 times out of 5, experience will save you from doing something expensively stupid. However once in a while it’s a clever generalist or someone who “doesn’t know any better” that uncovers a major advancement that the experienced people are blind to. So less than half of those failures are not thinking outside of the box, and most of the rest are not even knowing there was a box to begin with.

We tend to see this as an either or scenario. Throw a bunch of kids together or hire a bunch of cranky old men and women who say no all the time. That’s a false dichotomy. Hire some of both, don’t let the older ones steal all of the glory.

[colechristensen]

They just needed one farm kid who became an engineer.

They talked about taking seven weeks to figure out a conveyor, put me in the middle of Kansas and I could have told you how to solve that problem in seven minutes, give me a truck she a credit card and i can buy you the parts in an hour.

Bay Area engineering tends to be very sheltered, large groups of people with zero experience invading a domain. Sometimes it creates cool new solutions to old problems, other times it spends 100x the money on problems solved 100 years ago (And does a worse job at it too)

The real reason self driving cars will ultimately fail is they’re being designed and tested in places like Palo Alto where it doesn’t even rain half the year and a pothole is a thing of legend from lands far away.

[a_c]

On self driving car, I think we are solving all the problems caused by cars like air pollution and congestion with "better" car. First we have electric car, then we (will) have electric self driving car.

Public transport seems boring enough that no one talks about. Is there anything foundamentally wrong with buses and trains (and bike too) compared with cars? While we moved pass the faster horse era, we are now searching for better cars.

[lazide]

Once public transit is at the frequency of Tokyo trains and familiarity of the back of your hand, it’s often a no brainer.

The problem is, until it hits that point it involves a LOT of planning and conforming to other peoples schedules.

Which for many people (parents, folks going to and from a lot of different places) can dramatically decrease their ability to actually exist or do what they normally do.

If it’s in an environment where cars suck to drive/park/etc. (dense urban environments), then it quickly goes to public transit.

But especially in the US, density is low, cars are easier than the alternative unless someone spends several trillions at least on some giant initiative (which they won’t). So for most of the (physical) US, public transit makes no sense.

There are areas on the border (SF Bay Area, a lot of outlying Seattle Metro, etc), and in those cases both options suck.

[dublinben]

The real EV transportation revolution isn't electrifying cars, but micromobility like electric bikes, scooters, skateboards, etc. The fundamental inefficiency of trying to move a single person using two tons of machinery can never beat human scale transportation.

This doesn't even get into the enormous throughput efficiency that proper transit can deliver over longer distances.

[lazide]

Which works fine when you’re not trying to get through 6 inches of slush, 100+ F temps, or dealing with daily monsoon rains.

Which almost every part of the county except SF (pretty much) has to deal with some combination of regularly.

In a dense urban environment? Sure maybe, as long as the sidewalks and streets aren’t packed.

[colechristensen]

>Is there anything foundamentally wrong with buses and trains

It's slower, a round trip I could make in 30 minutes in a car would take careful scheduling and two hours on a bus. Sometimes it's cold outside and I'd have to walk half a mile to and from a bus stop. The local transit authority is experimenting with reducing the number of light rail cars so they'll be better covered by transit police in cars because violence is a problem. Once on BART I'm convinced somebody shit in a paper bag and left it in the middle of the car. Caltrain in rush hour would be a violation of the Geneva Convention if you made prisoners stand in such a cramped and unstable train for an hour at a time. I keep hearing about people getting stabbed by the local bus stop, 45 minutes after a friend visiting me came through the last time. A couple of months ago a dude died in the hospital after getting stabbed on a bus three blocks away from my apartment in a dispute over who got a cigarette left on the floor of the bus.

I calculated total cost of ownership of my car and it was considerably less than taking public transit unless I used public transit a lot, and any savings would be completely blown away by using Uber for just a couple of trips a month.

There are situations where public transit is a good thing, but it's generally a dirty, crime-infested, unpleasant experience in America only good for people who are really excited about not having a car. I'm opposed to any measures that force public transit on people by making car ownership more difficult until after they make public transit a safe, convenient, comfortable experience.

[Panzer04]

The way to make PT work is for it to be faster than cars. People should choose it because it’s faster/cheaper/more convenient, not because they have no alternative.

You’ll find this is typically the case in large cities that prioritise it appropriately - Between spending an hour in a car or on a train I’d choose the train every time.

[com2kid]

> It's slower, a round trip I could make in 30 minutes in a car would take careful scheduling and two hours on a bus.

That isn't a fundamental problem with buses, that is a problem with your local implementation.

A proper mass transit system gets to avoid all or most street traffic, including signaling and such.

Traveling through the streets of a city, you are lucky if your average speed hits 15mph. Mass transit can easily beat that if implemented well.

[mixmastamyk]

The problem here is your dysfunctional city, and has little to do with trains. I hear you however, mine is too.

[spc476]

In my area (southeast Florida), trains: not enough rail. Both trains and buses: not often enough. Buses: not enough routes, too many transfers (from experience, what would take 20 minutes by car took 1.5 hours by bus). Bikes: even though we have bike lanes, we get drivers from everywhere (tourist destination) and I wouldn't want to risk my life.

[galdosdi]

1) Public transport works today. True self driving is an unknown amount of time away, and IMHO it is a foolish mistake to assume it is definitely very soon or definitely very long before it works. It could be a year away or decades.

2) Even with self driving, there would be some efficiency gains from a bus, especially in places dense enough that there just isn't space for a car per person, like Manhattan. It would take adding something like 50 lanes of highway across the Hudson River to replace the capacity that trains currently provide for people to enter Manhattan from the mainland, IIRC. Neither electricity nor self driving in any way reduce the amount of space a car takes up. In fact self driving increases traffic by making it easier to go for a drive -- perhaps even send a car out for a 0 passenger cargo pickup trip.

Public transit is here to stay.

[oogabooga13]

I can agree with your general time scales on self driving. Just today in SF I watched firsthand a "Cruise" self driving car completely wig out when faced with a double parked car (pointed in the same direction as the "Cruise") and oncoming traffic on the opposite side of the street.

When it decided to make its manuever (on coming traffic briefly stopped to allow the car to drive around the double parked vehicle) the car made erratic micro turns and short hard braking action (pushing the nose of the car down) once it entered the (stopped) on coming traffic lane. The occupants were definitely thrashed around a bit.

The attempt was not pretty and definitely not even close to human level proficeny.

A typical manuever one has to make these days in the Bay with street parking eliminated in many busy restaurant / cafe corridors.

[bluGill]

I talk about public transit. I hate driving even though I do a lot of it (well not now that I work from home)

Transit can be great, but most of it is so bad nobody sane would use it. Why would you risk waiting for a bus that only comes twice an hour when your car is sitting in your driveway waiting for you? Why would you take a bus when you can walk almost as fast?

Both of those are very real problems that most transit has, and most people just ignore it.

[hinkley]

Taken cynically, these full disclosure engineering stories are calculated to make the team seem human and trustworthy.

How much do you want to bet that the real story is that one of them drove past a grain elevator and wondered if they make a portable version of that equipment, or one of the frustrated/amused locals didn’t suggest it. Either directly or subconsciously. One of my coworkers has spent most of the last year sharing observations and ideas with the team that are things I pointed out three years ago. I don’t know if he’s giving me an overly subtle nod or he just doesn’t know he’s doing it. Like people who accidentally write a story/song they heard once.

[corrral]

> One of my coworkers has spent most of the last year sharing observations and ideas with the team that are things I pointed out three years ago. I don’t know if he’s giving me an overly subtle nod or he just doesn’t know he’s doing it. Like people who accidentally write a story/song they heard once.

I once suggested a product name in a meeting and everyone was like "eh" and then about five minutes later someone else (no, not a manager) suggested the exact same name to instant acclaim, and it was adopted. I don't think she did it on purpose, so I wasn't and am not mad about it, but it was odd.

[hinkley]

Well that’s just egregious.

If the room reads right I’ll occasionally do that as a joke, but I have heard too many stories of people who do it for real (one of the reasons I try to make it a joke).

[orthecreedence]

Was she hot? I'm not being sarcastic, actually asking.

[colechristensen]

Clearly they have almost the right idea but missing just a bit of domain knowledge which would make the process incredibly easier.

[jfim]

If you want a laugh, you can always check out Zume pizza, which was this startup that made pizzas with robots.

They burned through several hundreds of millions in VC funding, even though a cursory look at their robots would show that they had no idea what they were doing. See for example https://www.bloomberg.com/news/articles/2020-02-13/inside-th...

[abracadaniel]

I imagine the frozen pizza industry has had that long figured out, and at scales far larger than they were hoping for.

[pkrein]

Want to come show us how it's done? :)

[colechristensen]

My email is my username @gmail.com I might be able to do some consulting.

[Raidion]

I think solving an "easy" problem is a good step on the way to a hard one, as long as the tool is set up to allow you to get out of a local maximum. Clearly there are locations where self driving is a smaller problem space than others, and I don't think it's bad to solve for those locations first.

I don't think it would be a fail if self driving cars were able to only operate in sunny locations with good roads. The caveat is that the outcome shouldn't just be a "magic" ML model that can't be modified to handle rain or potholes, it should be a set of tooling that allows you to make ML models that solve a variety of problems.

[munk-a]

So build a covered test road for your initial trials - then unleash them on real road situations. I'd be much more ready to trust a self-driving car that was born and raised on Boston streets than one in the wide, luxurious and pristine SoCal streets.

I'm not saying you need to teach your kid to swim by throwing them into the sea with weights tied around their ankles - you can ramp up to that... but you need to start exposing things to real world conditions pretty early on in development lest something, like a LIDAR censor close to the road surface in the front of the car, force you into a huge redesign when you discover that sheets of slush and road salt will liberally coat every front-facing surface of your vehicle driving in Boston in the winter.

[nebula8804]

You have literally described how Tesla(and the other Musk companies) have made both an excellent and terrible product at the same time. A big chunk of the Musk skeptic community consist of experienced professionals from all the industries that Elon is involved in. They are constantly bashing his companies for not paying attention to what has already been accomplished. Musk has had many failures due to ignoring the previously made mistakes that people before him have already made.

For example: Fit and finish, organizing your production process so you are not forced to build $50k cars in a tent in the parking lot, following proven practices for service etc.

At the same time, teardowns of the cars are showing a lot of out of the box thinking that other incumbent automakers are not following. Things such as eliminating redundant systems with newer designs that can do multiple tasks, making parts modular enough such that you don't have to wait for a new model year to implement, the megacastings etc.

In the end, Tesla has only survived because they had enough capital to make these new ideas stick and now others are starting to have to copy his company's accomplishments. So maybe as long as you have enough capital, then it may be worth to "relearn" everything that everyone else has learned because it may lead to asking "why" on some of the old ways of doing things.

[hinkley]

A similar story happened with digital cameras. A few of the film companies practically went out of business before the electronics got sorted out and what differentiated the good from the bad cameras circled back to the mechanical parts, which the incumbents had locked down ages ago. The pure digital play companies faded and some were acquired, and one day Nikon or Canon had the best DSLR system.

[nebula8804]

Thats unfortunate. What if there are certain aspects to existing cameras that are only there because a lesson was learned in the past but is no longer ideal in todays context. If there are no new comers that have the cash to burn to 're-evaluate' how everything is done today then we are missing out on potential innovations.

This process is like re-factoring old code but in real life.

[majormajor]

That's how Sony ate up a bunch of marketshare recently, actually, post-DSLR, by doing mirrorless early and well.

They were involved already in the industry as a component supplier, and then expanded to take advantage of the incumbents moving slowly.

[throwaway0a5e]

This is mostly a cultural problem.

Some professions will readily say "imma keep my mouth shut because this is not my area of expertise" whereas every software engineer is a subject matter on everything up to and including open ended and subjective questions that humans have been trying to conclusively answer all of history. (The various types of mechanical engineers are nearly as bad.)

One could argue that the cultural problem is just a reflection of a demographics problem but I don't care enough about that aspect to make that argument.

As an aside, the failure in TFA doesn't sound that dramatic. Seems like the age old tale of a customer running material handling equipment at the edge of it's comfortable range and an tech needing to be flown out to tell them where to slap vibrators on it or what other minor change needed to be made in order to make it work better. Yeah, they probably should have seen this coming and shipped a more flexible solution or at least have had shovel ready options that didn't involve flying engineers out but whatever, it's a prototype and this stuff happens.

[hinkley]

I belatedly figured out that my whole style of learning is based entirely on model building and testing, similar to what scientists do when they are exploring a new phenomenon. I keep meaning to look around for research on whether this is common to all autodidacts, just one mode, or if I’m rare.

The upshot of this is that I seem to be able to avoid the Expert Beginner trap and get into more of an Expert Journeyman, which is somewhat more useful and less dangerous. I’m often deputized to take over things I don’t actually know all that well, because I’m seen as having some knack for playing twenty questions and then being able to improvise reasonably well without running back to the delegator every fifteen minutes, or setting the building on fire for fear of asking for clarification when it’s warranted.

I am also pretty mechanically inclined, took a lot of things apart and back together, including but not limited to bicycles I’ve subsequently taken above 40 mph (like cars, there are many kinds of defects that only show up at 2x “normal” speed, because forces tend to quadruple). Software people who know hardware of some sort are, at least in my experience, generally safer about trying to defy the laws of physics.

But the danger with model testing is that I often sound like I know exactly what I’m talking about when I don’t (and sound too similar when I actually am the expert, so people either trust me too much or not enough). I’ve played with various levels of wiggle words and uncertain phrasing to try to fix this, always with mixed results. Sometimes even stating it as an educated guess, based on X and Y, causes other people to agree that sounds perfectly correct even if it’s not (one of the original definitions of a meme). Sometimes I avoid that trap of not believing my own model in my head, even if I don’t let on, sometimes I don’t. I should probably have more care about others aping my demeanor, but I tend to mentor anyone who is comfortable asking clarifying questions. It’s the ones that want to fake it til they make it that I can’t help.

The danger is, as always, in believing your own PR. Questioning it constantly is paralyzing and exhausting, both for yourself and for observers (especially the exhausting part). Questioning it not at all is exquisitely dangerous.

[bluGill]

If you are in software you have to be an expert in some other domain. Sure there are a few os and compiler writers who can sit in pure software land (even they need to know some about hardware). Pretty much everything else is for some other domain and so you need to know it.

[iancmceachern]

And change/difference for the sake of change/difference

[hinkley]

Everything in moderation. I do occasionally argue to let someone change something that doesn’t really need to be changed because it makes them happy. If they’re happy they’ll stay longer, or at least be less cranky the next time some random top-tier bullshit happens. Throw stress at someone whose fun you’ve completely sucked away and see how far that gets you.

The trick is to find something they can fiddle with without ruining everyone else’s fun in the process. Sometimes that’s easy, sometimes it’s a riddle.